We’ve all heard that “a picture is worth a thousand words”. That definitely applies to our hotel price comparison service hotel.idealo.de, a daughter company of Axel Springer. Each hotel supplies us with dozens of images, presenting us with the challenge of choosing the most “attractive” image for each pitch on our offer comparison pages since photos can be just as important for bookings as reviews. Given millions of hotel offers, we end up with more than 100 million images that require an “attractiveness” assessment.
Although teaching and training a computer to decide what makes a “beautiful” hotel photo is a hard problem, it’s not impossible. In this talk, we will present how we solved this difficult problem. In particular, we will share our training approaches and the peculiarities of the models. We will also show the “little tricks” that were key to solving this problem.
Key Takeaways:
- Explore the lifecycle of a large-scale deep learning project, from prototyping on a public dataset to fine-tuning on in-house labelled data to the deployment of the system in production and its business impact
- Learn a state-of-the-art approach to train neural networks with ordered labels (the Earth Mover’s Distance versus cross-entropy loss) and visualization techniques for CNNs
Check out the below video to know more about the talk.
